Introduction

Alkali-activated geopolymers (AAG) are generally known as novel hydraulic binder which can be prepared by alkali-activated aluminosilicate powders to replace Ordinary Portland Cement (OPC) in the manufacturing of engineering structures [1,2,3,4,5,6,7,8]. As a possible alternative application, it has been demonstrated that entrapment into a geopolymeric network of many hazardous compounds from inorganic waste or polluted products [9,10,11] can limit their release; in this context, they can be used to recycle fly ashes [12,13,14,15,16,17,18] or inertize some industrial by-products [19,20,21,22,23,24,25] thus showing a possible way which could permit to avoid their landfill disposal.

Dolomite refractories are often used in many steel production plants; in particular, they are used in the construction of basic oxide (BOF) and electric arc furnaces (EAF), but also in the building of ladle or other plants’ hot parts. At the end of their life, waste refractory products are generally disposed of in landfill due to their high amount of CaO- and MgO-containing compounds. In fact, their chemical composition causes volumetric instabilities which are a great disadvantage when exhaust dolomite refractories are used as a raw material for possible recycling. In fact, it is generally known that free CaO and MgO transform into hydroxides and carbonates under environmental conditions. In addition, CaO- or MgO-containing compounds suffer of retarded hydration which, in general, occurs when materials produced are in use; for example, when recycling concerns the production of mortars or concretes, the resulting hardened materials could suffer of progressive mechanical decay. The above phenomena could be amplified when CaO- and MgO-containing compounds are not in the form of fine powders, but their feature consists of granulated products. As a consequence, their use in the production of cement-based materials or in the manufacturing of roads or streets must be limited to very little quantities or require severe treatments before recycling as it has been clearly demonstrated by several research papers [26,27,28,29,30,31,32,33,34].

In order to avoid landfill disposal, in the present research, the possibility to recycle waste dolomite refractories into the production of geopolymer-based materials has been explored. The influence of particles size distribution of waste refractories on the performances of the resulting materials was studied.

The thermodilatometric behavior of the resulting materials was also investigated in order to evaluate the temperature range for their possible application.

Experimental Procedures

Starting Materials

Argical M1200 calcined kaolin (metakaolin-MK) received from Imerys Refractory Minerals was used as a SiO2 and Al2O3 source for the geopolymeric network formation. Such powder was mixed with a commercial sodium silicate solution, containing 58 wt.% of water with 2.23 wt.% SiO2/Na2O ratio (2.29 molar). Such characteristics are typical of generally available commercial products of moderate cost. The above mixture was activated by an 8 M NaOH solution prepared using sodium hydroxide pellets 98% (Titolchimica Spa-Pontecchio Polesine, Ro-IT). Such products were mixed in the required proportion following the route described in a previous paper [35]; the resulted hardened geopolymer, having the Si/Al, Al2O3/Na2O, and H2O/Na2O ratios of 1.63, 1.5, and 15.3, respectively, was used as the reference composition also in the present research.

The above blank composition was then added with different proportions of waste dolomite refractories. Such waste materials were received in the form of a granulated, wet product containing 11 wt.% of water. The presence of water is due to a previous weathering which was made by the producer to ease transport to its final landfill destination. The as-received material was dried in an oven at 100 °C for 24 h, transformed into a coarse powder by milling, and then shared into three parts: the first containing particles with maximum size of 250 μm, the second and the third with particles with maximum sizes of 1.0 mm and 2.5 mm, respectively. Each batch was used to prepare a complete set of geopolymer-based materials.

Chemical composition obtained by means of a Spectro Mass 2000 Induced Coupled Plasma (ICP) mass spectrometer and density determined following the EN993-1 standard of the as-received waste dolomite refractory are reported in Table 1.

Table 1 Chemical composition, reported in term of oxides, of the as-received waste dolomite refractory product
Full size table

The cumulative particle size distribution (PSD) of the product with maximum particles size of 2.5 mm was made by means of standard sieves, whereas those of the powders with maximum particles size of 1 mm and 250 μm, respectively, were acquired in water by an Horiba LA950 laser scattering particle size analyzer. Curves are displayed with logarithmic abscissa.

X-ray diffraction patterns, collected by a step size of 0.02° in the range 10°–80° and counting period of 40 s per angular abscissa, were obtained by a Philips X’Pert diffractometer at 40 kV and 40 mA using a Cu-Kα (Ni filtered) radiation. X’Pert HighScore software (Philips) was used for phase identification [36].

ATR-FTIR analysis was performed on the as-received materials as well as on hardened geopolymeric samples to investigate their chemical composition. A Nicolet iS™ 50 FT-IR Fourier transform spectrometer was employed using the Attenuated Total Reflection (ATR) method. Each IR spectrum was acquired using a 2 cm−1 resolution in a spectral window ranging from 500 to 4000 cm−1 with 100 scans.

Samples Preparation

The mix proportion design, referred to 100 g of MK, is reported in Table 2 together with sample identification symbolic names. Table also reports H2O/total solid content ratios of the materials prepared. Mixing was performed by a Hobart stirrer (5 L capacity). It must be also remarked that, for each composition (namely R100, R200, and R300) were prepared samples containing powdered waste refractory with different maximum particles dimension.

Table 2 Mix proportion design of the compositions prepared referred to 100 g of metakaolin (MK)
Full size table

The production protocol, after several attempts, was set up as follows: in the first step, the basic geopolymeric paste was produced and mixing was protracted until reaching a sufficient fluidity (15 min mixing); at that time the powdered refractory was added in the required proportion and mixing was maintained for 15 additional minutes. At the end of homogenization, slurries were poured into high-density cylindrical moulds, vibrated for 2 min in order to maximize air removal, and sealed by a plastic film.

In order to evaluate workability, the following modified spread-flow test has been pointed out [37, 38]: slurries were poured into a truncated conical mould (top diameter = 70 mm, bottom diameter r = 100 mm, height = 60 mm), which was filled up to the top; it was then lifted vertically after 1 min and the diameter of the paste after spreading was measured along two perpendicular directions. The slump value was derived from the average of the two measurements.

After casting, geopolymeric pastes were maintained into the moulds at room temperature for 24 h and then cured at 60 °C for further 24 h. Successively, they were removed from the moulds, sealed again by a plastic film, and maintained at room temperature and pressure for 7 days before being used for characterization and testing.

Characterization of Hardened Materials

Material's compressive strength was determined on cylindrical specimens with 60 mm diameter and 120 mm height with a crosshead speed of 2 mm min−1; data were averaged over five measurements for each composition in agreement with the ASTM C39 standard; measurements were performed by an 810 Material Test System.

Open porosity of the hardened samples was accessed by water absorption measurements. Investigations were made following a modified ASTM C 642 standard procedure on cylindric samples with diameter and height of 20 mm. In order to perform this tests, hardened samples were first dried at 80 ± 5 °C for 24 h and weighed in air. This measurement leads to determine samples dry weight (W1); in a second time, each sample was treated in an autoclave at 120 °C and 2 kPa for 2 h using 2 L of water, cooled down to room temperature (in water), dried with a cloth and weighed again: this measurements leads to determine its wetted weight (W2). Material’s water absorption, which in turn can be assumed also as open porosity, was evaluated according to the following equation:

$$ {\text{W}}\left( \% \right)\;\, = \;\,{1}00 \, \left( {{\text{W}}_{{2}} \; - \;{\text{W}}_{{1}} } \right)/{\text{W}}_{{1}}. $$

In accordance to the BET method, a Tristar 3000 gas adsorption analyzer (Micromeritics) was used to investigate specific surface areas and pore size distribution of the hardened materials on parts resulted from compressive strength tests; tests were made by nitrogen adsorption at 77°K.

Thermodilatometric tests were carried out in air, by a Netzsch alumina dilatometer, at a heating rate of 10 °C min−1 up to the softening temperature on cylindric samples (Ф = 3 mm and length = 25 mm). For this set of experiments, due to the small diameter of the samples, only geopolymers containing powders with maximum size of 250 μm were tested. During dilatometric tests, at the end of the shrinkage, samples collapse, and at that moment, tests were stopped. The collapsing temperature was assumed as the material’s softening temperature. The part of the thermodilatometric diagrams, relative to the softening temperatures, is not reported in the present paper because, in order to preserve dilatometer integrity from the aggression of the liquid, the experiment was stopped a few degrees before the final temperature.

Results and Discussion

The characterization of the as-received MK has been already described in a previous paper [35] and not repeated here. It should be remarked the presence of muscovite (K2Al4(Si6Al2O20)(OH)4) and anatase (TiO2) as additional phases to the amorphous metakaolinite. On the other hand, the present manuscript reports chemical and crystallographic characteristics of the waste dolomite refractories (R). As reported in Table 1, the main components of R are MgO and CaO which are the typical major compounds of dolomite refractories, Fe2O3 and MnO represent impurities deriving from steel production, whereas SiO2, Al2O3, and P2O5 are reasonably due to the binders which have been used for manufacturing the hot parts of steel production plants, i.e., refractory walls [39, 40].

The XRD investigation (Fig. 1), in agreement with data of Table 1, confirmed the presence of free periclase (MgO, PDF 01-087-0651), portlandite (Ca(OH)2, PDF 01-084-1276), calcite (CaCO3, PDF 01-086-0174), mullite (3Al2O3 2SiO2, PDF 01-079-1456), cordierite (Mg2Al4Si5O18, PDF 01-087-1964), and aluminum phosphate (Al(PO3)3, PDF 01-070-0104) together with some minor quantities of other hydrated compounds such as brushite (CaHPO4·2(H2O), PDF 00-011-0293) and pyroaurite (Mg6Fe2CO3(OH)16·4H2O, PDF 01-086-0181). The pattern displayed in Fig. 1 has been obtained on powders with maximum particles size of 250 μm, however, the two slots of powders with greater particles showed similar feature thus revealing an almost homogeneous crystallographic composition of the three batches of powders.

Fig. 1
figure 1

X-ray diffraction pattern of the as-received powdered waste dolomite refractory. Phases are identified with the following symbols: (open circle) = MgO; (open triangle) = Ca(OH)2; (closed triangle) = Ca(CO)3; (closed square) = 3Al2O32SiO2; (closed diamond) = Mg2Al4Si5O18; (open square) = CaHPO4·2(H2O); (open diamond) = Al(PO3)3; (closed circle) = Mg6Fe2CO3(OH)164H2O

Full size image

Figure 2a, b shows the cumulative particle size distribution trend of the three batches of powders. Figure 2a shows the results obtained on the product with maximum particles size of 2.5 mm and was obtained by standard sieves, whereas Fig. 2b displays the curves of the two batches of powders with maximum particles size of 250 (plain line) and 1000 μm (dashed line), respectively, and was obtained by the laser particles size analyzer. It can be observed that the powder with maximum size of 2.5 mm contains about 20% of particles having size lower than 250 μm and about 40% of particles with size greater than 1000 μm; on the other hand, the d50 of the powders with maximum size of 250 μm and 1000 μm are 125 and 435 μm, respectively.

Fig. 2
figure 2

a, b Particle size distribution (cumulative representation) of the three batches of powdered waste refractory: a obtained on the product with maximum particles size of 2.5 mm; b curves of the batches with maximum particles size of 250 (plain line) and 1000 μm (dashed line), respectively

Full size image

The mix proportion design of the materials prepared in the present research, reported in Table 2, shows that the addition of R lowers the H2O/total solid content of the different compositions. This ratio affects pastes workability, independently on the particles size distribution of the R powders used for materials preparation as confirmed by Fig. 3, which shows that the addition of R reduces the slump flow. In particular, it can be observed that the slurry corresponding to the blank composition has a slump flow of 155 mm; this value reaches the minimum level in the compositions of R300 which show a value close to the diameter of the truncated cone used for this test. Therefore, R300 has been established as the composition with the workability limit. Geopolymer-based materials containing greater quantity of R than the composition R300 were not prepared because components homogenization and pastes casting were not possible with the instruments available in our laboratories.

Fig. 3
figure 3

Slump flow versus materials composition: the plain line is referred to pastes containing the R with maximum size of 250 μm, whereas dashed and dotted lines are referred to the pastes made with R having maximum size of 1000 and 2500 μm, respectively

Full size image

Figure 4 shows three hardened cylindrical specimens with composition R300 made with R having particles of 250, 1000, and 2500 μm, respectively. In particular, it is observed that sample containing R particles with maximum size of 250 μm looks undamaged, the one made with R particles of 1000 µm shows the presence of some chipping (indicated by arrows), whereas that containing R particles of 2500 μm exhibits extended fractures [26, 30, 32, 33]. Therefore, the set of samples made with damaged products were not furtherly characterized. It can also be observed that all these three specimens show the presence of a significant amount of coarse open porosity which is reasonably caused by their low ratio of H2O/total solid content and low slump that limits pastes workability thus entrapping a large quantity of air during homogenization.

Fig. 4
figure 4

Image of the samples with composition R300 made with the three batches of powdered waste refractory. Detailing, from left to right, R with maximum size of 250, 1000, and 2500 μm, respectively. Arrows indicate some visible chipping

Full size image

Figure 5 compares the XRD patterns acquired on the mixture of powders containing 50 wt.% of MK and 50 wt.% of R before geopolymerization (top) with the one acquired on the geopolymer with composition R300 (bottom). Such comparison enables evaluation of the possible partial reaction of some starting components during the geopolymer production.

Fig. 5
figure 5

X-ray diffraction patterns, acquired on the mixture of powders containing 50% wt of MK and 50% wt of R (i.e., R300), before (top) and after (bottom) geopolymerization. Phases are identified with the following symbols: (closed star  = SiO2 (quartz); (closed plus) = (K2Al4(Si6Al2O20)(OH)4) (muscovite); (closed circle) = Mg6Fe2CO3(OH)164H2O; (open square) = CaHPO4·2(H2O); (open triangle) = Ca(OH)2; (open circle) = MgO; (closed square) = 3Al2O32SiO2; (closed triangle) = Ca(CO3); (open diamond) = Al(PO3)3

Full size image

It has been observed that the intensity of peaks at 9.0°, 17.6°, and 19.8° (corresponding to the muscovite phase of MK) appear sensibly reduced in the geopolymer with respect to the unreacted powders mixture. This is in agreement with literature [7, 12, 16,17,18, 20,21,22,23, 35] which clearly state the reactivity of silica and most of aluminosilicate compounds during geopolymerization.

Reduced intensities after the geopolymer production are observed in peaks from pyroaurite (11.3° and 22.7°), brushite (11.7° and 23.4°), portlandite (18.1°, 34.1°), and periclase (42.9° and 62.4°) thus revealing that also these compounds are prone to react during the geopolymer production. On the other hand, quartz and calcium carbonate seem to behave as inert components as their peaks maintain similar intensity before and after geopolymerization.

In fact, literature clearly reports that the presence of MgO or CaO into a geopolymeric network could have beneficial effects on the properties of the hardened materials because their reactions limit material’s shrinkage and refine the pore size distribution within the geopolymeric paste provided that the components used for materials preparation are in the form of fine powders [41,42,43,44,45,46,47,48,49]. However, it is reasonably expected that the presence of such compounds into coarse particles, being cause of uncontrolled instabilities, could generate fractures and chipping in hardened materials as it has been observed also in cement-based mortars containing coarse steel slag and therefore coarse CaO- and MgO-containing particles [32,33,34].

Figure 6 compares ATR-FTIR spectra from a geopolymer reference (black curve) with R300 sample before and after polymerization (blue and red curve, respectively). The band between 950 and 1200 cm−1 is attributed to Si–O–Si vibrations [16, 50,51,52,53]. The spectral position of this peak is strictly connected to the geopolymerization process. The chemical reaction leads to a shift toward lower wavenumbers in both “Reference” and “R300 Post-polymerization” samples. [16, 50,51,52,53]. Peaks at 1640 cm−1 and 3200 cm−1 are connected to the presence of O–H bonds in adsorbed water [50, 51]. The weak bands at ~ 870 cm−1 and ~ 1450 cm−1 are related to the presence of carbonates [54]. The analysis of the FTIR spectra confirms XRD results and provides further insights about the chemistry of such geopolymers. In particular, the addition of refractories does not affect the chemical composition of the final geopolymer.

Fig. 6
figure 6

ATR-FTIR spectra collected on a geopolymer reference, on a R300 sample before and after geopolymerization. The main vibrations are indicated by arrows

Full size image

Figure 7 shows the trend of compressive strength versus materials composition of the samples made with R having maximum size of 250 μm (plain line) and 1000 μm (dashed line), respectively (error bars are also reported). It can be observed that both set of materials display the highest value at composition R100 and that higher R addition lower strength; in addition all samples prepared using R with maximum size of 250 μm display higher strength compared to the control samples and to those containing R of greater size as it is already demonstrated by other authors [55].

Fig. 7
figure 7

Compressive strength versus composition of the two sets of samples made with R having maximum size of 250 μm (plain line) and 1000 μm (dashed line), respectively (error bars are also reported)

Full size image

Figure 8 reports water absorption versus composition of the samples prepared. The plain line refers to materials prepared with R powders having maximum size of 250 µm, whereas the dashed line refers to those obtained using R powders with maximum size of 1000 µm. The R-free control composition revealed water absorption of 24% which decreases in each set of materials for composition R100; higher R additions cause the increase of water absorption. However, the set of materials containing powders with maximum size of 250 μm revealed values that remain inferior to one of the reference composition for any R addition, whereas those produced using powders with maximum size of 1000 μm showed greater values.

Fig. 8
figure 8

Water absorption versus composition of the two sets of samples made with R having maximum sizes of 250 μm (plain line) and 1000 μm (dashed line), respectively

Full size image

Water absorption tests provide information about material's open porosity, i.e., great pores and defects which affect their mechanical performances. Nevertheless, water adsorption tests are useful to evaluate the homogeneity of the bulk material. It should be pointed out that samples with composition R100 are those with the highest strength and the lowest water absorption.

Figure 9 and Table 3 show the results obtained from pore size distribution and specific surface area investigations. It may be observed that all specific surface area data fall around 25 m2 g−1 and seem to be not affected by materials composition, whereas pore volume increases little with the addition of R. In parallel, pore size distribution of all the compositions range between the same limits (5 and 55 nm) and shows maximum pores concentration at about 21 nm for all compositions; the quantity of pores having this size, increases with the addition of R.

Fig. 9
figure 9

Pore size distribution of geopolymeric samples with composition R0 (black), R100 (red), R200 (green), and R300 (yellow) (color figure online)

Full size image
Table 3 Specific surface area (m2 g−1) and pore volume (cm3 g−1) of geopolymeric samples made with dolomite waste powder with maximum size of 250 μm
Full size table

However, material’s mechanical behavior is related, but not intrinsically linked to the nanosized porosity when the microstructure also contains a large quantity of micro and macroporosity. In our experimental investigation, the relation between mechanical strength and materials water absorption appears much more evident since it is observed as a great correlation between compressive strength and water absorption data.

Figure 10 is a stereoscopic light microscope photograph of a polished section of a sample with composition R300; it highlights the presence of several rounded holes which are reasonably due to air bubbles entrapped during production. It has been moreover observed that their amount appears to increase with the amount of R and therefore be directly related to pastes workability.

Fig. 10
figure 10

Stereoscopic light microscope photograph of a polished section of a sample with composition R300; several rounded holes are visible

Full size image

It is therefore possible to speculate that materials’ nanoporosity plays a partial role on materials’ mechanical behavior. It is the opinion of the authors of the present research that materials’ macroporosity, i.e., the one which includes micro and millimetric porosity plays a much more important role. This type of porosity is due to the different water/total solid content of the starting compositions which affects their workability and therefore their air content during production and hardening [33, 38, 56,57,58], in this way showing that vibration is not sufficient to remove all the air entrapped into the pastes.

Figure 11 shows the semi-quantitative thermodilatometric behavior of the four specimens (i.e., R0, R100, R200, and R300), whereas the characteristic temperatures obtained during the thermodilatometric tests are reported in Table 4. It is observed that all compositions display an almost continuous shrinkage with increasing temperature, but display a significant length change between 170 and 400 °C. In this range, it has been observed a shrinkage of about 0.1% independently on materials composition. The real sintering process starts at higher temperature (above 600 °C for all compositions) and then, in a restricted range, samples collapse and melt at temperatures ranging from 650 to 750 °C as a function of their composition. It must be remarked that the amount of CaO and MgO, generally known as blasting compounds, which is improved by enlarging the quantity of R into the geopolymers, have little effect on the sample’s melting temperature, prevailing the influence of the geopolymeric matrix. This is reasonably due to the particular chemical composition of the geopolymeric matrix which is placed in the area of the ternary diagram Al2O3–SiO2–Na2O where are present three ternary eutectic temperatures, respectively, 760, 732, and 740 °C [59]. The gap between the above eutectics and the softening temperatures measured in the present research is due to the non-equilibrium conditions of the thermodilatometric tests which therefore reveal higher values but it is clear that the addition of waste dolomite refractory reduces refractoriness of the geopolymeric blank matrix. The above results are partially in conflict with those obtained by other researchers who demonstrated that alkali-activated geopolymers may have a thermal stability up to 1000 °C [60,61,62,63]. The difference is reasonably due to the different chemical composition, namely the Si/Al, Na2O/Al2O3 ratios of the geopolymer investigated.

Fig. 11
figure 11

Thermodilatometric behavior of geopolymeric samples prepared with waste dolomite refractory powder having maximum size of 250 μm

Full size image
Table 4 Samples’ shrinkages and softening temperature measured during the thermodilatometric tests
Full size table

Figure 12 shows the changes occurring in FTIR spectra on the reference sample after exposure to high temperatures. The detail in panel (b) highlights the intensity decreasing in the band located at 3250 cm−1. This band arises from to the O–H bonds from water which are removed after heating and no other crystallographic changes were observed. For the sake of brevity, the FTIR spectra of the other compositions are not reported in the present paper, but material's dehydroxylation is the only crystallographic change which has been observed, in the present research, as a consequence of the thermal treatment up to 500 °C in line with the results obtained by other authors [64, 65].

Fig. 12
figure 12

ATR-FTIR spectra collected on a reference sample before and after heating at 500 °C (a). Details of a spectral windows ranging from 1400 to 4000 cm−1 (b). The main vibrations are indicated by arrows

Full size image

Conclusions

Metakaolin, waste dolomite refractories, sodium silicate solution, and 8 M sodium hydroxide solution were mixed in order to prepare several geopolymers. The as-received waste refractories were previously crumbled and then transformed into powdered products which were shared into batches having three maximum particles dimensions namely 250 μm, 1 mm. and 2.5 mm. Each batch was added, in different proportions, to the geopolymeric matrix.

It was observed that

  • The addition of powdered waste refractory reduces slurries workability but, after hardening, improves the compressive strength with respect to the reference composition;

  • Samples containing particles with maximum sizes of 1 mm and 2.5 mm do not maintain integrity after hardening and they are fractured as a consequence of significant volumetric instabilities;

  • During the geopolymeric reaction quartz (present in MK) and calcium carbonate (from waste refractory) behave as inert components;

  • Muscovite (from MK), pyroaurite, brushite, portlandite, and periclase (from waste refractory) display a certain degree of reactivity thus disclosing their contribute to the build up of the geopolymeric network;

  • The highest compressive strength was measured in composition R100, whereas higher or lower R additions lead to materials with inferior strength;

  • Specific surface area of all the samples seems to be independent of materials composition, whereas pore volume slightly increases with the amount of waste dolomite refractory;

  • During the thermodilatometric tests, all compositions display a shrinkage of about 0.1% between 170 and 400 °C, but sintering starts at higher temperature (above 600 °C) and samples melt at temperatures ranging from 650 to 750 °C as a function of their composition.